|
|
vol. 3, nr. 2 (2004)
|
Soil profile thermal conditions evaluation by means of artificial neural networks
|
|
|
Paweł Licznar1, Marian Rojek2
|
|
| (get PDF )
|
|
|
1 Institute of Building and Landscape Architecture, University of Agriculture, Pl. Grunwaldzki 24, 50-363 Wrocław |
|
|
2 2Institute for Land Reclamation and Environmental Development, University of Agriculture, Pl. Grunwaldzki 24, 50-363 Wrocław |
|
|
vol. 3 (2004), nr. 2,
pp. 317-323
|
|
|
abstract:
Soil temperature is an important factor limiting agricultural use of the land. Thermal conditions are significant for seeds' germination and the former plants' growth. Presently the soil temperature's profiles are being subject of the detailed observation conducted on the automatic meteorological stations. Artificial neural networks implementation for the soil profile thermal conditions analysis was the aim of the research. The study was made on the base of the air and soil temperature records registered in the first half-year of 2000 at the Wrocław-Swojec meteorological station. Five perceptrons with single hidden layer of five neurons were developed and applied for prediction of soil temperature at different depths on the base of the minimal air temperature 5 cm above the ground and the soil temperature measured at the chosen depths. The results of the study showed that the neural networks may be successfully used for supplementing broken soil temperature measurement series. Moreover they provide a possibility to limit the number of conducted observations, necessary for the whole soil profile thermal conditions evaluation, to only two parameters: the minimal air temperature 5 cm above the ground and the soil temperature at the depth of 5 cm or 50 cm.
|
|
keywords:
soil temperature measurements, artificial neural networks, automatic meteorological stations
|
|
original in:
Polish
|
|
|
|
|
|
